#clojure log - Aug 21 2011

The Joy of Clojure
Main Clojure site
Google Group
List of all logged dates

0:35 * srid just realized that battling with clj-http is likely unnecessary as it may not work on appengine, that it is probably better to use appengine's http client itself.

0:36 ibdknox: srid: that is probably true

0:43 srid: reading the java examples to do this makes me cringe, http://stackoverflow.com/questions/637100/

0:43 nothing like a python one-liner ... urlopen(url).read(). sigh.

0:47 ibdknox: since that's a buffer I think you can just slurp it

0:48 I didn't look closely though

0:51 srid: slurp returns java.lang.String. the URL in question actually returns gzip encoded data. so slurp 'decoded' a gzipped binary stream?!

0:52 right, as expected i cannot decompress the data returned by slurp: jvm throws "Not in GZIP format"

0:52 ibdknox: ah, like I said, I didn't look at it closely ;)

0:53 isn't there a header you can pass to say you don't support gzip?

0:53 that would be the simple solution

0:53 srid: no, the site will always return gzip compressed data; here's their rationale - http://api.stackoverflow.com/1.1/usage/gzip

0:54 ibdknox: wow

0:54 that makes me a little sad

1:24 srid: damn, I don't know why this error keeps coming up when compiling a .clj file:

1:24 java.lang.ClassCastException: java.lang.String cannot be cast to clojure.lang.IObj

1:24 this is what I have in the top of the file:

1:24 (ns notaskinnerbox.stackexchange

1:24 (:require [clj-http.client :as client])

1:24 (:import [java.util.zip GZIPInputStream] [java.net URL]))

1:26 or the error is happening somewhere else, but line number is always 1

1:29 yup, it was on a different line. compiler error reporting is less than ideal.

3:04 furd: I'm having issues getting lein to do... anything really

3:04 If I create a new lein project and go into it

3:04 and attempt to open the repl or get deps or do any commands

3:04 It gives me https://gist.github.com/1160272

3:10 Any ideas why? I'm not finding anything via google

3:14 Running any lein command, even help or version, causes that error in a project directory

3:17 jli: furd: looks like something's wrong with your project.clj

3:18 can you put it somewhere?

3:18 this is from looking at "Caused by: java.lang.ClassCastException: java.lang.Integer cannot be cast to clojure.lang.Named (project.clj:1)"

3:18 furd: yeah it's from a brand new lein new command, here I'll put it on the gist

3:19 Added it as a comment, though it removed the indentation for some reason

3:20 Oh is it as my project name is an int

3:20 pmbauer: furd: (defproject 14 <---

3:20 14 is not a symbol, but a numeric literal

3:20 So you can't name a lein project 14

3:21 furd: Never would have assumed you can't name your project a number, though I suppose aside from doing Project Euler I don't know why you would

3:21 pmbauer: furd: try 'lein new fourteen'

3:22 or euler14

3:22 furd: Will do, thank you very much.

7:47 msappler: hey

7:48 trying out amazon cloudfront for my clojure game application, can somebody test please (java webstart should work linux/mac/win): http://resatori.com/cyber-dungeon-quest

7:51 mikera: @msappler - works for me.... looks good! what graphics library are you using and how are you wrapping it?

7:53 msappler: I am using slick2d

7:54 i have open sourced a previous alpha you can take a look herehttps://code.google.com/p/clojure-rpg/

8:05 MasseR: msappler: Works great

8:08 A bit sluggish on loading and starting though

8:12 msappler: thanks ok

8:20 MasseR: And by "a bit sluggish" I mean that it took like 5 minutes starting the game after loading the jar files

8:31 msappler: yeah i know it takes a bit long

8:32 maybe because so many class files

8:33 michaelr525: Hi!

8:33 Is there a standard way to dechunkify sequences?

8:34 I have a sequence of URLs that I intend to scrap one by one..

8:35 mrBliss`: michaelr525: http://blog.fogus.me/2010/01/22/de-chunkifying-sequences-in-clojure/

8:36 michaelr525: thanks, I've seen this one

8:36 Was wondering maybe there is something more standard

8:49 fantazo: hi, could someone say me, what I'm making here wrong, so that the innerest fn of "connect-with" doesn't get executed? from my syntactic and semantic understanding of clojure it should execute. but why doesn't it? http://pastebin.com/0gqumjd0

8:55 raek: fantazo: 'map' and 'for' are lazy and generate a sequence of values, which are only calculated when they are used. you are probably looking for (dorun (map ...)) or (doseq ...) if you want the side-effects from the functions rather their return values

9:00 fantazo: also: (swap! atom (fn [x] (foo x a b c))) --> (swap! atom foo a b c) (map (fn [x] ...) coll) --> (for [x coll] ...)

9:38 fantazo: raek, thanks.

9:49 when I want something like: {:a 1 :b {:thing (1 2 3)}} => {:a 1 :b {:thing (1 2 3 4}}, what is the best routine in the clojure library to do it? I'm currently trying to plug it together by myself, but it seems that I'm doing it not the clojure way.

9:53 raek: fantazo: ##(update-in {:a 1, :b {:c [1 2 3]}} [:b :c] conj 4)

9:53 lazybot: ⇒ {:a 1, :b {:c [1 2 3 4]}}

9:54 michaelr525: how do you people debug in emacs?

9:54 i'm not sure this sldb thing is working..

9:54 fantazo: raek, you are very helpful, thank you sir.

9:55 michaelr525: I get an exception and want the check the parameters of the function which has thrown the exception

9:55 to check

10:05 it shows no locals

10:05 why?

10:05 clojurebot: why not?

10:05 michaelr525: becase

10:11 raek: michaelr525: on the JVM the locals are not stored when an exception is thrown, so at the time you get hold of it they're gone. this is a bit unfortunate, but is the reason why that part of the slime/swank protocol is not implemented for clojure

10:11 you need to use a real debugger to be able to see the locals

10:13 for pure functions debugging is pretty simple: just call them with the interesting parameters in the repl. for non-pure code, the need for a debugger is of course more obvious

10:15 so my personal way of debugging is to write as much of the code (that makes sense to be functional) in a functional style

10:16 if a function does a computation and writes the result, split it into two functions: one that only does the calculation (pure) and one that does the I/O (impure)

10:24 pdk: sounds about what haskell imposes usually

10:30 Norrit: Hi, any idea where I can find the source code of let* ?

10:31 mudge: Is there a way in clojure to cause destructuring to work like this: (let [{keys}] code) and have each value in the keys collection assigned to a name of the same key for each value ?

10:32 Norrit: mudge: (let [{:keys [fred ethel lucy]} m] ...

10:33 mudge: Norrit: yes, i want it to work like that except I don't want to list the vector ie [fred ethel lucy], how do i do the same thing but not list the vector?

10:33 Norrit: so the vector is implicitly the keys in the collection

10:34 Norrit: ah, I don't think that is possible at the moment

10:35 perhaps you could write a macro

10:36 mudge: Norrit: yea, maybe

10:57 michaelr525: hey people

10:57 so... can anyone help me with slime debugging?

10:58 I'm using clojure-jack-in and I don't get locals in stack traces when an exception is throws

10:58 thrown

10:58 is this a known issue?

11:00 mudge: I want to user clj-time but when I look in Clojars, there's multiple jars for it: http://clojars.org/search?q=clj-time

11:01 how do i know which clj-time I should use?

11:01 how do I find out the latest and best clj-time to use?

11:03 michaelr525: mudge: I'd look for the original author latest version and continue from there

11:06 mudge: the original author doesn't have clj-time on Clojars, or at least i can't tell if he does or not

11:09 michaelr525: then take the latest version

11:19 raek: michaelr525: locals in stack traces are not suported without a debugger, for the reasons I mentioned before

11:22 michaelr525: raek: sorry, I wasn't connected, is it in the channel log?

11:23 raek: should I use cdt for debugging then?

12:43 jli: /a/u

12:43 whoops

12:50 srid: learning clojure makes me humble; there is much to learn.

12:58 dnolen: hmm how can you do a fast array type instance check?

13:04 oh yes matching on primitive array is disgustingly fast, https://gist.github.com/1160849

13:06 jli: dnolen: nice. what's the :vec for?

13:06 dnolen: jli: :vec is a vector pattern

13:06 that means you're pattern matching on a datatype that supports random access

13:07 :vec without extra specialization works for persistent vectors

13:09 but you can specialize the pattern matching to other types like i've shown here, with little effort

13:09 not certain on the syntax yet for this, but it's working.

13:15 hmm for primitive arrays I suppose you'd like to specify an offset … then you can use recur without much hassle

13:18 saalaa__: could anyone give me a hand on this:

13:18 (defmacro object [identifier documentation fields] `(defstruct ~identifier ~(keys fields)))

13:18 as you'll probably have noticed, the struct won't be created correctly because the result of (keys fields) is a list

13:19 I can't seem to find a way to "break" that list

13:19 dnolen: saalaa__: defstruct is deprecated. use ~@ to splice in the keys.

13:19 saalaa__: dnolen: oh right

13:20 also I'm using an old version of clojure

13:20 thanks, I'll give ~@ a try

13:21 I don't really need the features of defstruct; I guess I'll go for a map

13:45 dnolen: matching bits, a bit silly but fun, https://gist.github.com/1160849

13:46 amalloy: mudge: you can't do the destrucuring thing you wanted

13:47 (let [{[k1 k2]} m] ...) would be illegal because a map has to have an even number of entries: keys and values

15:18 dnolen: well match leaves Scala in the dust when matching primitive arrays. > 30X faster.

15:33 faster than Racket too...

15:41 mudge: when I run "lein uberjar jobboard", it create two files, a jar file and a file named jobboard, what is the jobboard file?

15:42 why does "lein uberjar jobboard" create two files?

16:05 srid: question regarding accessing clojure docs: where do I go to see a list of all functions that are specific to a clojure hash map?

16:06 for eg., I was specifically looking for a function that would return a subset of the map (chosen keys specified by me)

16:08 ibdknox: ,(doc select-keys)

16:08 clojurebot: "([map keyseq]); Returns a map containing only those entries in map whose key is in keys"

16:08 ibdknox: http://clojure.org/data_structures

16:08 srid

16:09 srid: right, thanks! .. i missed the official site.

16:16 coopernurse: srid: clojuredocs is also good, and has a quick ref that groups by data structure http://clojuredocs.org/quickref/Clojure%20Core

16:16 I like the examples on clojuredocs

16:26 srid: coopernurse: that's very handy! examples are what I most like about clojuredocs as well.

16:26 coopernurse: btw, when I clicked on your link I see "Logged in as coopernurse" on the top-right corner.

16:26 session caching bug?

16:26 coopernurse: srid: woah. on the clojuredocs site it says that?

16:26 srid: yes

16:27 I hit the same bug while developing code.activestate.com

16:27 then I remembered the I mistakenly cached the page header too. so if one user loaded the page, and when the next did the same immediately he would see the former user's page header

16:27 coopernurse: interesting.. so this is just a cosmetic issue?

16:27 for example, can you change my email address?

16:28 on my clojuredocs profile page

16:28 srid: no, it is only a page caching bug. it just loads it from the cache for me (which cache was last-updated when you did)

16:29 coopernurse: ah, interesting. I'll see if anyone has reported that on their feedback site

16:30 I don't see anything related to that

16:30 * srid reported https://github.com/zkim/clojuredocs/issues/35

16:31 coopernurse: perfect, thanks

16:31 srid: I see you're at ActiveState.. I've walked by those offices many times (I'm in Seattle)

16:32 srid: oh, nice.

16:35 michaelr525: hey people!

16:44 ibdknox: Open call for feedback on Noir: https://groups.google.com/forum/#!topic/clj-noir/rh3f8CyR9RE

16:44 I'm hoping to get 1.2.0 out in the next few days

16:45 srid: would creating a appengine noir project be as simple as running 'lein noir new' (without having to custom patch things later)?

16:46 (in 1.2.0 rel., I mean)

16:48 coopernurse: srid: not sure. I've used appengine-magic to make appengine compojure projects

16:48 but not noir

16:49 https://github.com/gcv/appengine-magic

16:50 dnolen: new blog post, Solving the Mapping Dilemma - http://dosync.posterous.com/solving-the-mapping-dilemma

16:50 please upvote on HN if you feel inclined.

16:52 srid: weird, (println) doesn't print anything for some reason. prints on repl, but not via 'lein run'

16:54 coopernurse: dnolen: great stuff

16:54 dnolen: coopernurse: thx!

16:55 eskatrem: is 4clojure.com down today?

16:57 michaelr525: so, nobody ever has a problem with chunking? or everybody is using fogus' solution? or maybe there is some other secret dechunking trick? :)

16:58 srid: chunking? not sure what you are talking about.

16:59 michaelr525: chunking in sequences..

16:59 srid: ah ok

16:59 ibdknox: srid: there's a post in the mailing list about getting Noir running on appengine

16:59 dnolen: michaelr525: you can force one at time

17:00 michaelr525: dnolen: how?

17:00 ibdknox: srid: https://groups.google.com/d/topic/clj-noir/5a-3mYzs8dA/discussion

17:00 michaelr525: dnolen: oh, btw I like your blog posts but I think you are a bit crazy ;)

17:01 ibdknox: all the good ones *are* crazy

17:01 (inc dnolen)

17:01 lazybot: ⟹ 3

17:01 dnolen: (fn lazier [s] (when s (cons (first s) (lazy-seq (lazier (next s))))

17:01 michaelr525: ^

17:02 michaelr525: let me try it..

17:02 it should go into core

17:03 dnolen: michaelr525: most of people that ask for it are solving Project Euler ;)

17:04 ibdknox: hmm

17:04 is clojurescript broken for anyone else if you do script/bootstrap

17:04 I just did a fresh clone of the repo

17:05 michaelr525: me not. I'm just trying to scrap some website. I have a lazy list of urls, when evaluated the urls are actually fetched. so instead of fetching them one by one, I fetch them 24 by 24

17:07 skelternet: Should I expect to have to resort to a macro when implementing a with-someresource pattern that uses a var?

17:08 srid: how do find the root cause of an error when the traceback has no source/line information? eg: https://gist.github.com/1161172

17:08 (inc srid)

17:08 lazybot: You can't adjust your own karma.

17:08 ibdknox: fun.

17:09 if you set clojurescript home with a trailing slash it blows up

17:09 lol

17:09 srid: ,(loop [u "srid"] (inc u) (recur u))

17:09 clojurebot: #<ClassCastException java.lang.ClassCastException: java.lang.String cannot be cast to java.lang.Number>

17:09 * ibdknox sighs

17:09 joshuahickman: Was there another intended effect of that code?

17:10 ibdknox: he's trying to increment his own karma

17:10 by getting the bot to do it

17:10 tufflax: ,(println "(inc srid)")

17:10 clojurebot: (inc srid)

17:10 lazybot: ⟹ 1

17:10 joshuahickman: lol

17:10 srid: wow, nice

17:10 tufflax: hax!

17:10 srid: i was trying to find an exploit

17:10 tufflax: &(println "(inc srid)")

17:10 lazybot: ⇒ (inc srid) nil

17:10 ibdknox: OMG you're ruining our karma system

17:10 lol

17:11 tufflax: ,(println "(dec srid)")

17:11 clojurebot: (dec srid)

17:11 lazybot: ⟹ 0

17:11 tufflax: :)

17:11 clojurebot: Huh?

17:11 ibdknox: hah

17:12 * srid dreams of self-evolving bots competing for karma-resources and evolving into more intelligent bots

17:14 amalloy: eskatrem: yikes, thanks for the heads up. i'll see if i can figure out what's going oin

17:16 srid: answering my own question (above) - i forgot to put a "&" to form: (let [[a & b] aList] ...)

17:16 eskatrem: amalloy: I cant believe I was the only one using it! that website is awesome

17:16 tufflax: been there, done that, got the t-shirt

17:16 srid: tracebacks are not very useful, i have to think in possibilities to find where the error could have come from.

17:17 would be very nice to have an IDE (including emacs) that can directly highlight the sexp form where the exception originates from

17:17 amalloy: eskatrem: no idea what was wrong, but the server was not running at all. back up now

17:18 eskatrem: amalloy: thank you. I am still stuck in the "easy" category, but I think I am improving

17:19 michaelr525: srid: i wish i had locals in my tracebacks..

17:19 srid: i would very happy even if there was line number information.

17:21 joshuahickman: You know, Dr. Scheme's error tracebacks are really useful

17:21 amalloy: dnolen: isn't that just (defn lazier [s] (iterate rest s))?

17:21 joshuahickman: I wish more IDEs were like that, particular ones where I don't have to use scheme

17:21 raek: srid: use (require ...ns... :reload) to get line numbers for the definitions. while in the stack trace, press "v" when an item is selected to jump to that place

17:21 amalloy: dnolen: or i guess rest knows about and preserves chunks, never mind

17:23 raek: you don't get line numbers if you load code by sending it to the repl, but you get them if clojure loads the code for you.

17:23 srid: raek: i don't understand. in the stack trace I saw <https://gist.github.com/1161172> there were no symbols used in my source, except for some unusual references like "core.clj:2382" (but core.clj has only a dozen lines or so)

17:23 raek: the "v" thing might not work for everything

17:24 ah, this is a compiler exception...

17:24 those are indeed not very helpful

17:25 amalloy: eskatrem: weird. the logs indicate that someone was looking up /problems/63—

17:25 raek: srid: the core.clj is clojure/core.clj

17:25 michaelr525: dnolen: this lazier func did something weird

17:25 raek: srid: which version of clojure are you using?

17:25 amalloy: afaict it just issued a warning and survived, but what a bizarre uri

17:25 michaelr525: maybe I should not defn it?

17:25 srid: raek: 1.2.1 I believe.

17:26 raek: srid: do you have a paste of the code?

17:26 it is a syntax error somewhere in the code, but the exception message is not very helpful for finding out where

17:26 michaelr525: it evaluated the lazy seq inside a defn while compiling from emacs-slime

17:26 this is really bizarre

17:27 dnolen: michaelr525: paste

17:27 eskatrem: amalloy: I was looking at "Drop every Nth item"

17:27 raek: ,(let [[a b c] {:a 1, :b 2}] nil)

17:27 clojurebot: #<UnsupportedOperationException java.lang.UnsupportedOperationException: nth not supported on this type: PersistentArrayMap>

17:27 srid: raek: yes, the error was caused my forgetting to include a "&" in the loop assign [item & items] and [idx & indices]; see the full diff here, https://github.com/srid/notaskinnerbox/commit/db9bcd7b444b5985d6607242b8189e82f8f8ae06

17:28 michaelr525: dnolen: ok. I will prepare a small test case

17:28 srid: the commit above was the fixed version with & properly included.

17:29 raek: it seemed more of a runtime error (because destructuring happens at runtime) than syntax error (shouldn't syntax errors be caught during compile time?)

17:30 amalloy_: srid: you can write invalid destructuring forms

17:31 &(let [{x :x}] (inc x)) ;; compiler exception

17:31 lazybot: java.lang.IllegalArgumentException: let requires an even number of forms in binding vector

17:32 raek: &(fn [] (let [{x :x}] (inc x)))

17:32 lazybot: java.lang.IllegalArgumentException: let requires an even number of forms in binding vector

17:32 raek: &(fn [] (let [[x y z] {:a 1, :b 2}] nil))

17:32 lazybot: ⇒ #<sandbox13213$eval17885$fn__17886 sandbox13213$eval17885$fn__17886@7792e0>

17:32 raek: srid: you are right. it was a runtime exeption...

17:34 hrm. and the stacktrace includes the line "at notaskinnerbox.core$_main.doInvoke(core.clj:24)"

17:34 coopernurse: I want to run (read-string) against each line in a file I'm reading, but still treat it as a lazy sequence. any suggestions?

17:34 raek: which points to the line which you fixed

17:35 coopernurse: I'm using (line-seq rdr) - and basically want to wrap (read-string) around it

17:35 raek: coopernurse: then you don't want to slurp the file and call read-string on that

17:35 you can call read repeatedly on the reader

17:35 coopernurse: raek: ok thanks

17:35 raek: to lazilly read each form

17:35 coopernurse: but I thought (line-seq) is lazy

17:36 raek: this requires you to store the data as separate top-level forms, though

17:36 srid: raek: ah. I overlooked that. yes, #24 is the error's origin.

17:36 raek: coopernurse: ah, if each line is a clojure form then that works fine too

17:36 srid: yet the traceback _can_ be improved a bit.

17:36 coopernurse: yep, each line is a separate form

17:37 raek: then (map read-string (line-seq ...)) should work too

17:37 coopernurse: ah, of course

17:37 thanks

17:37 I keep forgetting that map is lazy

17:38 joshuahickman: I've got an insane question

17:38 Is there any way to get leiningen to work for vanilla projects?

17:38 It seems the people at work like it significantly better than their current build system

17:38 coopernurse: raek: that worked great, thanks

17:39 joshuahickman: *vanilla JAVA

17:39 I suppose that vital detail should be included

17:39 raek: it should work for simple projects, at least

17:40 joshuahickman: Do you know if there is a reference for this somewhere?

17:40 Google isn't helpful

17:40 coopernurse: does lein just wrap maven calls?

17:40 raek: "For projects that include some Java code, you can set the :java-source-path key in project.clj to a directory containing Java files. (You can set it to "src" to keep Java alongside Clojure source or keep them it in a separate directory.) Then the javac compiler will run before your Clojure code is AOT-compiled, or you can run it manually with the javac task."

17:40 coopernurse: oh, nice

17:40 michaelr525: dnolen: here http://pastie.org/2408368

17:40 coopernurse: that could come in handy

17:40 raek: joshuahickman: see the tutorial and the :java-source-path option

17:41 someone mentioned a problem where lib/ was not on the classpath for javac, but I dunno if that has been solved

17:41 joshuahickman: Hmm...

17:41 Thanks

17:42 raek: I just thought you might want to know that. if it is a bug, then it will probably be fixed very soon (updates are released quite frequently)

17:43 dnolen: michaelr525: so what's the problem?

17:44 joshuahickman: Yeah, that's ligit

17:45 I'll have to figure out if having more than one leiningen project is needed, and if so, how to set up dependencies between them in that case

17:46 dnolen: gotta run.

17:47 michaelr525: hmm

17:47 thanks anyway

17:51 coopernurse: joshuahickman: you could probably use a standard maven repo to store your dependencies between projects. I haven't read the docs to see if you can tell lein which repo to publish to

17:52 yep, looks like you can. see this thread: http://stackoverflow.com/questions/3468461/push-to-nexus-using-leiningen

17:52 so you could run your own maven repo internally (e.g. artifactory) and setup your lein projects to deploy to that

17:54 going to head out for a while too.. bye all

18:02 eskatrem: why is this line not working? (apply and [true true false])

18:03 gstamp: and is a macro

18:03 eskatrem: and it's not possible to apply a macro over a vector? how to do that then?

18:04 gstamp: I guess you could try something like this: (reduce #(and %1 %2) [false true true])

18:05 not sure if there is a better alternative

18:07 lucca: and normally takes a &rest

18:07 well.. that sort of thing

18:07 not &rest here, heh

18:07 if you reduce, you can't get the short-circuit eval

18:08 gstamp: that's true

18:08 (excuse the pun)

18:09 eskatrem: (reduce #(and %1 %2) [false true true]) works

18:09 lucca: that was already mentioned

18:11 the difficulty here is you're mixing strict eval functions and a non-strict eval macro

18:11 something with reductions might be a better approach...

18:29 michaelr525: damn

18:30 I can't get slime/swank to show locals in exception traces

18:30 I'm trying to do it from noon and now it's 1:30 am

18:31 hhh

18:33 raek: michaelr525: it can't.

18:33 you need a debugger for that

18:33 http://pastebin.com/FzWEXR9j

18:35 mudge: hey, anybody here use lein ring uberwar ?

18:36 i used lein ring uberwar to drop my clojure program into tomcat

18:36 and tomcat has been deploying my war file for the last 20 minutes

18:36 20 minutes seems too long for tomcat to deploy a file, usually it takes 20 seconds

18:36 any thoughts?

18:41 michaelr525: raek: but I am using a debugger, or at least I think so. check cdt: http://georgejahad.com/clojure/swank-cdt.html, and here is a video that demonstrates some debugging http://www.youtube.com/watch?v=d_L51ID36w4

18:41 ibdknox: mudge: I've done it

18:41 it was instantaneous

18:42 do you have any code that runs outside of a function?

18:55 mudge: hi ibdknox

18:55 ibdknox, no why?

18:56 ibdknox: that's often the source of such weird errors

18:56 for example

18:56 you can have code that starts up some thread (like in a def) and then sit forever waiting for you project to jar :)

18:57 mudge: ibdknox: ah okay,

18:58 yea, basically I have my tomcat6 trying to deploy my war file and nothing happens

18:59 the tomcat log says: INFO: Deploying web application archive jobs.war and it is deploying forever

18:59 and doesn't stop deploying

19:05 ibdknox: weird

19:05 I've definitely seen it work

19:05 but I've only done it once

19:05 lol

19:07 mudge: i restarted tomcat and my war file is deployed now

20:06 replaca: 0.

20:46 chewbranca: anyone know of a readability port in clojure? all my google searches return flame wars about lisp readable-ness :/

20:47 I was keeping an eye on goose, which was a java lib, but apparently they ported over to scala

20:49 found this: https://github.com/getwoven/webmine taking a peek now

22:11 dsantiago: dnolen, I took a look at match today and couldn't figure out how regexes would get added from the source code. Would it be something you add to LiteralPattern?

22:11 duncanm: dnolen: nice blog post

22:11 dnolen: duncanm: thx

22:12 dsantiago: no, you would need to add a RegexPattern datatype.

22:13 dsantiago: it's probably a bit too early days to try and do this yourself :) I need to write up a detailed explanation about how to extend match.

22:13 duncanm: dnolen: having match with regexp would be cool

22:13 dnolen: i liked that about scala, but i was a bit disappointed that they don't statically determine the number of capture groups

22:14 dnolen: duncanm: it would, how to implement RegexPattern yourself would probably be a good match tutorial. but I'm a bit busy with VectorPattern at the moment...

22:15 duncanm: interesting, I'll have to remember that when I take a shot at getting it working in match.

22:15 dsantiago: dnolen, yeah, I was not filled with confidence after taking a peek.

22:15 But it would be really awesome.

22:16 dnolen: dsantiago: the part that's a bit of mind warp is manipulating the matrix. that needs serious step-by-step explanation.

22:18 dsantiago: Yeah, I wasn't too clear on any of it, even before that.

22:18 The p-for-n stuff, or whatever that was.

22:19 icey: ibdknox: the remote stuff in pinot is pretty cool. Is there anything on the horizon for raising events from the server to clients? (i.e. push / comet?)

22:19 dnolen: dsantiago: you're a brave person for even daring to take a peek :) there's a lot renaming in match's future.

22:23 dsantiago: You can always hope it's just a really easy oversight.

22:24 icey: it looks like closure has something meant for bi-directional server/client communication called BrowserChannel; but it appears to need appropriate handlers running on the server to make it work

22:28 dnolen: dsantiago: I'll probably be adding - match.array, match.bits, match.regex soon so definitely keep your eyes open. should be easier to understand how to do it when they are not mixed in with everything else.

22:31 dsantiago: dnolen, why is it that the array manipulations aren't general?

22:32 dnolen: dsantiago: what do you mean?

22:32 dsantiago: I guess I don't understand why every type has to rewrite its own version of the matrix manipulation.

22:33 dnolen: dsantiago: vector / array / byte all share the same matrix manipulation.

22:33 map and sequences are fundamentally different datastructures so must be deal w/ differently.

22:34 dsantiago: But aren't we talking about an array of regexes here?

22:34 Array or vector?

22:34 Seems like the true/false logic from string/int equality would apply.

22:35 dnolen: I'm assuming people will want to match the seq of matches produced by the regex tho right?

22:35 dsantiago: Oh, hm. That's not what I was thinking about.

22:37 I was thinking like you give it a vector of strings, ints, whatever, and if the literal is a regex, it has to match a string.

22:37 So like, you could break a web route on /'s, and then match it against a vector of strings and regexes.

22:38 dnolen: you could definitely do that.

22:40 dsantiago: Without having to do the matrix specialization? :)

22:40 dnolen: dsantiago: if you wanted to keep it simple, yes it wouldn't be much work, you wouldn't need manipulate the matrix.

22:40 you just need to implement RegexPattern and it's p-to-clj method should do the test.

22:41 dsantiago: Nice.

22:41 chewbranca: dnolen: dsantiago late to the conversation, ran into the same issue last night with regex and match, but wasn't sure how to add it in; luckily my problem allowed me to drop regex and instead simplify the match clause down to one piece and basically do the heavy lifting later

22:42 here's what I ended up doing, using a combination of match and multimethods: http://paste.lisp.org/display/124168

22:42 dsantiago: dnolen, which class should I be mimicking?

22:42 dnolen: dsantiago: LiteralPattern.

22:42 chewbranca: I'm still new to clojure and lisp, so I still get my mind blown how simple and elegant some things can be in it

22:42 dnolen: chewbranca: fun stuff :)

22:43 chewbranca: that code was ported from an object oriented hierarchy of processing classes

22:43 dnolen: chewbranca: nice you're taking advantage of the expression sugar.

22:43 chewbranca: much cleaner

22:43 dnolen: (match [(expr)] ...)

22:44 chewbranca: dnolen: yeah that was helpful, I was using expressions as the result of what to run on a match, but realized that would get ugly in a hurry, then realized multimethods would work very well with it

22:47 dnolen: so what I was trying to understand about match, is if its possible (or a likely feature) to allow expressions in the match clauses? as right now it seems only static elements are allowed as match clauses, and I could definitely see it getting tricky and slow in a hurry to allow everything to be expressions, but I don't see how you would do something like a regex in your match clauses otherwise?

22:48 dnolen: chewbranca: yes, I'm thinking about supporting apply functions to things being matched.

22:48 chewbranca: wow that was not a well structured sentence

22:48 dnolen: [(1 :app inc)]

22:48 so that would match only if after inc is applied.

22:49 chewbranca: dnolen: hrmm... not exactly following that, how would that work with a regex for example?

22:50 dnolen: for regexes it would just be something like

22:50 [(_ :regex #"hello")]

22:50 or

22:51 [("foo.com" :regex #"^www\.")]

22:51 chewbranca: it seems like the underlying logic matrices you use to determine what columns had the most relevance would be very tricky if you can't figure out what its operating on, or is that what you mean by using apply, where you would apply the regex to a particular column, so you would at least be able to score the columns

22:52 dnolen: chewbranca: not really, :app is a kind of pattern that can be scored.

22:53 [#"hello"] also valid of course, re: regex syntax.

22:53 chewbranca: dnolen: ok I see what you mean, so you declare the match specifically to be an operation operating on some column, I was confused how you would determine what the column would be if you're just using an expression

22:55 dnolen: chewbranca: yeah group by operation. putting really complex operations inside the match clauses is kinda gnarly anyhow.

22:55 complex expressions I mean.

22:56 chewbranca: very true

23:09 dnolen: chewbranca: dsantiago: minimal regex example.

23:09 https://github.com/swannodette/match/blob/vector-patterns/src/match/regex.clj

23:10 dsantiago: How funny, right as I hit the button to fork.

23:10 And you're already done.

23:11 dnolen: don't consider this the API tho, just something to play around with.

23:12 dsantiago: That's pretty much what I was planning to write.

23:12 chewbranca: dnolen: oh nice, that's slick

23:12 dsantiago: Pretty much RegexPattern exactly.

23:12 dnolen: chewbranca: yeah not much code eh :)

23:12 dsantiago: I'd have run into more trouble with the rest.

23:14 chewbranca: dnolen: yeah code quantity is impressive

23:15 dnolen: to String bit not even really necessary, that's just for debugging.

23:15 toString I mean.

23:16 check it out

23:16 https://github.com/swannodette/match/blob/vector-patterns/src/match/regex.clj

23:17 chewbranca: dsantiago: way shorter/cleaner if you use defrecord

23:17 clojurebot: llahna: anyway the answer is no. you can use #(some-fn %1 default-arg %2), for example

23:18 dsantiago: dnolen, looks good enough for me to use.

23:19 chewbranca: dnolen: oh interesting, I need to look more into deftype and defrecord, just started looking into them today

23:20 dnolen: dsantiago: you should just copy and paste that for now, don't really want to mainline that just yet.

23:20 dsantiago: OK.

23:21 chewbranca: dnolen: cool, just grabbed a copy of it

23:27 any of the getwoven guys hang out here?

23:39 technomancy: ibdknox: pong

23:49 chewbranca: technomancy: hey quick question, any plans on supporting direct git sources in lein projects, or just sticking with local maven repos?

23:51 technomancy: chewbranca: not sure if it's quite what you want, but that sounds a bit like checkout dependencies; see the faq

23:55 chewbranca: technomancy: checking it out

23:55 * talios has been thinking of adding something similar to the maven plugin, but also kinda seems prone to nastyness, and promoting broken artifacts :(

23:56 dnolen: ,(.indexOf [:a :b :c :d] :c)

23:56 clojurebot: 2

Logging service provided by n01se.net